Quote:
That is a very nice bit of creative writing. I especially like the vocabulary. It's like poetry. Amazing how he could memorize the sound of an amp he had never heard before and identify it by name with a break of a couple of minutes between samples. Almost unbelievable. No, not almost...
Assuming that the blind test is valid (it may or may not be), I wonder if the following:
"This momentary aberrant behavior seems to occur whenever there's a particularly loud high frequency signal during an instant of high-complexity, where the DAC has a lot of work to do to sort everything out. It's as if the ODAC just gives up, momentarily, and renders those instantaneous high frequencies like two blocks of styrofoam being rubbed together - to produce a very short in duration, but very loud distortion of information that the LX and the PB2 have no difficulty rendering cleanly."
is related to inter-sample clipping, as tested
here on two sound cards, one with a CS4398 DAC, and another with PCM1792A (ironically, it is the more expensive latter one that fails) ? With my test sample, which is a mix of a 11.025 kHz sine wave at an amplitude of 0.8284 and 45 degrees phase, and a 15.025 kHz sine wave at 0.4142 amplitude and 0 phase, the problem is plainly audible on a DAC that fails. However, it is a rather unrealistic "worst case" signal and a similar effect is much less likely to occur at an audible level in typical music. When I mentioned this issue to the designer of the ODAC, he argued that it would never be audible with real music; from this response, and the fact that clipping problems are not easy to avoid with the ES902x DAC chips (as evidenced by the NuForce uDAC2), I suspect that the ODAC would fail my test, too.
Then again, maybe there was some more trivial problem, like the USB power not meeting the specs, and the DAC clipping already below 0 dBFS. Or the input stage of the iBasso amplifier clipping the output of the ODAC, which was the loudest of the three DACs tested. In any case, the subjective description of the problem sounds like some sort of clipping issue to me.